A New Model for Mortgage Loss Given Default by Leveraging Time to Recovery

The research report titled “A New Model for Mortgage Loss Given Default by Leveraging Time to Recovery” delves into the intricacies of credit risk, specifically focusing on the Loss Given Default (LGD) model. Credit risk pertains to the potential financial loss a lender might face due to a borrower's inability to repay a loan. The study emphasizes the significance of three models: LGD, EAD (Exposure at Default), and PD (Probability of Default) in assessing credit risk. The LGD model, in particular, represents the portion of a loan that a lender stands to lose if a borrower defaults. An essential component of this model is the Time to Recovery (
$T$
2R), which indicates the duration required for a lender to recover the outstanding balance after a default event. The research incorporates T2R into the LGD model, addressing data censoring issues and offering a more comprehensive understanding of loss given default. Using a dataset of defaulted mortgage loans, the study identifies key influencing features and establishes correlations between LGD and T2R. The research further introduces a linear regression model and a survival model to estimate T2R. Finally, the study predicts the LGD across the sample dataset, distinguishing between open and closed episodes, highlighting the elevated risk in the open episode phase.